1 |
dpavlin |
1 |
/* |
2 |
|
|
* integrate Hyper Estraier into PostgreSQL |
3 |
|
|
* |
4 |
|
|
* Dobrica Pavlinusic <dpavlin@rot13.org> 2005-05-19 |
5 |
|
|
* |
6 |
|
|
* TODO: |
7 |
|
|
* - all |
8 |
|
|
* |
9 |
|
|
* NOTES: |
10 |
|
|
* - clear structures with memset to support hash indexes (who whould like |
11 |
|
|
* to create hash index on table returned from function?) |
12 |
|
|
* - number of returned rows is set by PostgreSQL evaluator, see: |
13 |
|
|
* http://archives.postgresql.org/pgsql-hackers/2005-02/msg00546.php |
14 |
|
|
* |
15 |
|
|
* Based on: |
16 |
|
|
* - C example from PostgreSQL documentation (BSD licence) |
17 |
dpavlin |
41 |
* - coreexample002.c and nodeexample002.c from Hyper Estraier (GPL) |
18 |
dpavlin |
1 |
* - _textin/_textout from pgcurl.c (LGPL) |
19 |
|
|
* |
20 |
|
|
* This code is licenced under GPL |
21 |
|
|
*/ |
22 |
|
|
|
23 |
|
|
#include "postgres.h" |
24 |
|
|
#include "fmgr.h" |
25 |
|
|
#include "funcapi.h" |
26 |
|
|
#include "utils/builtins.h" |
27 |
|
|
#include "utils/array.h" |
28 |
dpavlin |
56 |
#include "utils/lsyscache.h" |
29 |
dpavlin |
1 |
#include "miscadmin.h" |
30 |
dpavlin |
61 |
#include "commands/trigger.h" |
31 |
|
|
#include "executor/spi.h" |
32 |
|
|
|
33 |
dpavlin |
1 |
#include <estraier.h> |
34 |
|
|
#include <cabin.h> |
35 |
dpavlin |
41 |
#include <estnode.h> |
36 |
dpavlin |
1 |
|
37 |
|
|
#define _textin(str) DirectFunctionCall1(textin, CStringGetDatum(str)) |
38 |
|
|
#define _textout(str) DatumGetPointer(DirectFunctionCall1(textout, PointerGetDatum(str))) |
39 |
|
|
#define GET_STR(textp) DatumGetCString(DirectFunctionCall1(textout, PointerGetDatum(textp))) |
40 |
|
|
#define GET_TEXT(cstrp) DatumGetTextP(DirectFunctionCall1(textin, CStringGetDatum(cstrp))) |
41 |
|
|
|
42 |
dpavlin |
27 |
/* SortMem got renamed in PostgreSQL 8.0 */ |
43 |
|
|
#ifndef SortMem |
44 |
|
|
#define SortMem 16 * 1024 |
45 |
|
|
#endif |
46 |
|
|
|
47 |
dpavlin |
85 |
#ifdef PG_MODULE_MAGIC |
48 |
|
|
PG_MODULE_MAGIC; |
49 |
|
|
#endif |
50 |
|
|
|
51 |
dpavlin |
40 |
#define ATTR_DELIMITER "{{!}}" |
52 |
dpavlin |
57 |
#define HINTS_PREFIX "HINTS." |
53 |
dpavlin |
40 |
|
54 |
dpavlin |
5 |
/* prototype */ |
55 |
|
|
char *attr2text(ESTDOC *doc, char *attr); |
56 |
dpavlin |
41 |
char *node_attr2text(ESTRESDOC *rdoc, char *attr); |
57 |
dpavlin |
56 |
void cond_add_attr(ESTCOND *cond, char *attr); |
58 |
dpavlin |
1 |
|
59 |
|
|
|
60 |
|
|
/* work in progress */ |
61 |
dpavlin |
19 |
PG_FUNCTION_INFO_V1(pgest_attr); |
62 |
|
|
Datum pgest_attr(PG_FUNCTION_ARGS) |
63 |
dpavlin |
1 |
{ |
64 |
dpavlin |
31 |
ArrayType *attr_arr = PG_GETARG_ARRAYTYPE_P(6); |
65 |
dpavlin |
25 |
Oid attr_element_type = ARR_ELEMTYPE(attr_arr); |
66 |
|
|
int attr_ndims = ARR_NDIM(attr_arr); |
67 |
|
|
int *attr_dim_counts = ARR_DIMS(attr_arr); |
68 |
|
|
int *attr_dim_lower_bounds = ARR_LBOUND(attr_arr); |
69 |
dpavlin |
19 |
int ncols = 0; |
70 |
|
|
int nrows = 0; |
71 |
|
|
int indx[MAXDIM]; |
72 |
dpavlin |
25 |
int16 attr_len; |
73 |
|
|
bool attr_byval; |
74 |
|
|
char attr_align; |
75 |
dpavlin |
1 |
ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; |
76 |
|
|
AttInMetadata *attinmeta; |
77 |
|
|
TupleDesc tupdesc; |
78 |
dpavlin |
19 |
Tuplestorestate *tupstore = NULL; |
79 |
dpavlin |
1 |
HeapTuple tuple; |
80 |
|
|
MemoryContext per_query_ctx; |
81 |
|
|
MemoryContext oldcontext; |
82 |
|
|
Datum dvalue; |
83 |
|
|
char **values; |
84 |
dpavlin |
19 |
int rsinfo_ncols; |
85 |
dpavlin |
1 |
int i, j; |
86 |
dpavlin |
19 |
/* estvars */ |
87 |
dpavlin |
25 |
ESTDB *db; |
88 |
|
|
ESTCOND *cond; |
89 |
|
|
ESTDOC *doc; |
90 |
|
|
int ecode, *est_result, resnum; |
91 |
|
|
int limit = 0; |
92 |
|
|
int offset = 0; |
93 |
|
|
|
94 |
dpavlin |
19 |
char *index_path; |
95 |
|
|
char *query; |
96 |
|
|
char *attr; |
97 |
dpavlin |
31 |
char *order; |
98 |
dpavlin |
1 |
|
99 |
dpavlin |
19 |
|
100 |
|
|
/* only allow 1D input array */ |
101 |
dpavlin |
25 |
if (attr_ndims == 1) |
102 |
dpavlin |
19 |
{ |
103 |
dpavlin |
25 |
ncols = attr_dim_counts[0]; |
104 |
dpavlin |
19 |
} |
105 |
|
|
else |
106 |
|
|
ereport(ERROR, |
107 |
|
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
108 |
|
|
errmsg("invalid input array"), |
109 |
|
|
errdetail("Input array must have 1 dimension"))); |
110 |
|
|
|
111 |
dpavlin |
1 |
/* check to see if caller supports us returning a tuplestore */ |
112 |
|
|
if (!rsinfo || !(rsinfo->allowedModes & SFRM_Materialize)) |
113 |
|
|
ereport(ERROR, |
114 |
|
|
(errcode(ERRCODE_SYNTAX_ERROR), |
115 |
|
|
errmsg("materialize mode required, but it is not " \ |
116 |
|
|
"allowed in this context"))); |
117 |
|
|
|
118 |
dpavlin |
19 |
/* get info about element type needed to construct the array */ |
119 |
dpavlin |
25 |
get_typlenbyvalalign(attr_element_type, &attr_len, &attr_byval, &attr_align); |
120 |
dpavlin |
19 |
|
121 |
dpavlin |
1 |
/* get the requested return tuple description */ |
122 |
|
|
tupdesc = rsinfo->expectedDesc; |
123 |
dpavlin |
19 |
rsinfo_ncols = tupdesc->natts; |
124 |
dpavlin |
1 |
|
125 |
|
|
/* |
126 |
|
|
* The requested tuple description better match up with the array |
127 |
|
|
* we were given. |
128 |
|
|
*/ |
129 |
dpavlin |
19 |
if (rsinfo_ncols != ncols) |
130 |
|
|
ereport(ERROR, |
131 |
|
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
132 |
|
|
errmsg("invalid input array"), |
133 |
|
|
errdetail("Number of elements in array must match number of query specified columns."))); |
134 |
|
|
|
135 |
dpavlin |
1 |
/* OK, use it */ |
136 |
|
|
attinmeta = TupleDescGetAttInMetadata(tupdesc); |
137 |
|
|
|
138 |
|
|
/* Now go to work */ |
139 |
|
|
rsinfo->returnMode = SFRM_Materialize; |
140 |
|
|
|
141 |
|
|
per_query_ctx = fcinfo->flinfo->fn_mcxt; |
142 |
|
|
oldcontext = MemoryContextSwitchTo(per_query_ctx); |
143 |
|
|
|
144 |
|
|
/* initialize our tuplestore */ |
145 |
|
|
tupstore = tuplestore_begin_heap(true, false, SortMem); |
146 |
|
|
|
147 |
dpavlin |
19 |
|
148 |
|
|
/* take rest of arguments from function */ |
149 |
|
|
|
150 |
|
|
/* index path */ |
151 |
|
|
if (PG_ARGISNULL(0)) { |
152 |
|
|
ereport(ERROR, |
153 |
|
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
154 |
|
|
errmsg("index path can't be null"), |
155 |
|
|
errdetail("Index path must be valid full path to HyperEstraier index"))); |
156 |
|
|
} |
157 |
|
|
index_path = _textout(PG_GETARG_TEXT_P(0)); |
158 |
|
|
|
159 |
|
|
/* query string */ |
160 |
dpavlin |
31 |
if (PG_ARGISNULL(1)) { |
161 |
dpavlin |
19 |
query = ""; |
162 |
|
|
} else { |
163 |
|
|
query = _textout(PG_GETARG_TEXT_P(1)); |
164 |
|
|
} |
165 |
|
|
|
166 |
|
|
/* atribute filter */ |
167 |
|
|
if (PG_ARGISNULL(2)) { |
168 |
|
|
attr = ""; |
169 |
|
|
} else { |
170 |
|
|
attr = _textout(PG_GETARG_TEXT_P(2)); |
171 |
|
|
} |
172 |
dpavlin |
31 |
|
173 |
|
|
/* sort order */ |
174 |
|
|
if (PG_ARGISNULL(3)) { |
175 |
|
|
order = ""; |
176 |
|
|
} else { |
177 |
|
|
order = _textout(PG_GETARG_TEXT_P(3)); |
178 |
|
|
} |
179 |
dpavlin |
19 |
|
180 |
dpavlin |
31 |
|
181 |
dpavlin |
19 |
/* limit */ |
182 |
dpavlin |
31 |
if (PG_ARGISNULL(4)) { |
183 |
dpavlin |
19 |
limit = 0; |
184 |
|
|
} else { |
185 |
dpavlin |
31 |
limit = PG_GETARG_INT32(4); |
186 |
dpavlin |
19 |
} |
187 |
|
|
|
188 |
|
|
/* offset */ |
189 |
dpavlin |
31 |
if (PG_ARGISNULL(5)) { |
190 |
dpavlin |
19 |
offset = 0; |
191 |
|
|
} else { |
192 |
dpavlin |
31 |
offset = PG_GETARG_INT32(5); |
193 |
dpavlin |
19 |
} |
194 |
|
|
|
195 |
|
|
|
196 |
|
|
/* open the database */ |
197 |
|
|
elog(DEBUG1, "pgest_attr: est_db_open(%s)", index_path); |
198 |
|
|
|
199 |
|
|
if(!(db = est_db_open(index_path, ESTDBREADER, &ecode))){ |
200 |
|
|
ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
201 |
|
|
errmsg("est_db_open: can't open %s: %d", index_path, ecode), |
202 |
|
|
errdetail(est_err_msg(ecode)))); |
203 |
|
|
} |
204 |
|
|
|
205 |
dpavlin |
20 |
elog(DEBUG1, "pgest_attr: query[%s] attr[%s] limit %d offset %d", query, (PG_ARGISNULL(2) ? "NULL" : attr), limit, offset); |
206 |
dpavlin |
19 |
|
207 |
|
|
/* create a search condition object */ |
208 |
|
|
if (!(cond = est_cond_new())) { |
209 |
|
|
ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED), |
210 |
|
|
errmsg("pgest_attr: est_cond_new failed"))); |
211 |
|
|
} |
212 |
|
|
|
213 |
|
|
/* set the search phrase to the search condition object */ |
214 |
|
|
if (! PG_ARGISNULL(1) && strlen(query) > 0) |
215 |
|
|
est_cond_set_phrase(cond, query); |
216 |
|
|
|
217 |
|
|
/* minimum valid attribute length is 10: @a STREQ a */ |
218 |
|
|
if (! PG_ARGISNULL(2) && strlen(attr) >= 10) { |
219 |
dpavlin |
40 |
elog(DEBUG1,"attributes: %s", attr); |
220 |
dpavlin |
56 |
cond_add_attr(cond, attr); |
221 |
dpavlin |
19 |
} |
222 |
|
|
|
223 |
dpavlin |
31 |
/* set the search phrase to the search condition object */ |
224 |
|
|
if (! PG_ARGISNULL(3) && strlen(order) > 0) { |
225 |
|
|
elog(DEBUG1,"est_cond_set_order(%s)", order); |
226 |
|
|
est_cond_set_order(cond, order); |
227 |
|
|
} |
228 |
|
|
|
229 |
dpavlin |
38 |
if (limit) { |
230 |
|
|
elog(DEBUG1,"est_cond_set_max(%d)", limit + offset); |
231 |
|
|
est_cond_set_max(cond, limit + offset); |
232 |
|
|
} |
233 |
|
|
|
234 |
dpavlin |
19 |
/* get the result of search */ |
235 |
|
|
est_result = est_db_search(db, cond, &resnum, NULL); |
236 |
|
|
|
237 |
|
|
/* check if results exists */ |
238 |
|
|
if ( 0 == resnum ) { |
239 |
|
|
elog(INFO, "pgest_attr: no results for: %s", query ); |
240 |
|
|
} |
241 |
|
|
|
242 |
|
|
/* total number of tuples to be returned */ |
243 |
|
|
if (limit && limit < resnum) { |
244 |
dpavlin |
31 |
nrows = limit; |
245 |
dpavlin |
19 |
} else { |
246 |
|
|
nrows = resnum - offset; |
247 |
|
|
} |
248 |
|
|
|
249 |
|
|
|
250 |
|
|
elog(DEBUG1, "pgest_attr: found %d hits for %s", resnum, query); |
251 |
|
|
|
252 |
dpavlin |
1 |
values = (char **) palloc(ncols * sizeof(char *)); |
253 |
|
|
|
254 |
|
|
for (i = 0; i < nrows; i++) |
255 |
|
|
{ |
256 |
dpavlin |
19 |
|
257 |
|
|
/* get result from estraier */ |
258 |
|
|
if (! ( doc = est_db_get_doc(db, est_result[i + offset], 0)) ) { |
259 |
dpavlin |
47 |
elog(INFO, "pgest_attr: can't find result %d", i + offset); |
260 |
dpavlin |
19 |
} else { |
261 |
|
|
elog(DEBUG1, "URI: %s\n Title: %s\n", |
262 |
|
|
est_doc_attr(doc, "@uri"), |
263 |
|
|
est_doc_attr(doc, "@title") |
264 |
|
|
); |
265 |
|
|
} |
266 |
|
|
|
267 |
|
|
/* iterate over results */ |
268 |
dpavlin |
1 |
for (j = 0; j < ncols; j++) |
269 |
|
|
{ |
270 |
dpavlin |
19 |
bool isnull; |
271 |
|
|
|
272 |
|
|
/* array value of this position */ |
273 |
dpavlin |
25 |
indx[0] = j + attr_dim_lower_bounds[0]; |
274 |
dpavlin |
19 |
|
275 |
dpavlin |
25 |
dvalue = array_ref(attr_arr, attr_ndims, indx, -1, attr_len, attr_byval, attr_align, &isnull); |
276 |
dpavlin |
19 |
|
277 |
|
|
if (!isnull && doc) |
278 |
|
|
values[j] = DatumGetCString( |
279 |
|
|
attr2text(doc, |
280 |
dpavlin |
20 |
(char *)DirectFunctionCall1(textout, dvalue) |
281 |
dpavlin |
19 |
)); |
282 |
|
|
else |
283 |
|
|
values[j] = NULL; |
284 |
dpavlin |
1 |
} |
285 |
|
|
/* construct the tuple */ |
286 |
|
|
tuple = BuildTupleFromCStrings(attinmeta, values); |
287 |
|
|
|
288 |
|
|
/* now store it */ |
289 |
|
|
tuplestore_puttuple(tupstore, tuple); |
290 |
dpavlin |
19 |
|
291 |
|
|
/* delete estraier document object */ |
292 |
dpavlin |
47 |
if (doc) est_doc_delete(doc); |
293 |
dpavlin |
1 |
} |
294 |
|
|
|
295 |
|
|
tuplestore_donestoring(tupstore); |
296 |
|
|
rsinfo->setResult = tupstore; |
297 |
|
|
|
298 |
|
|
/* |
299 |
|
|
* SFRM_Materialize mode expects us to return a NULL Datum. The actual |
300 |
|
|
* tuples are in our tuplestore and passed back through |
301 |
|
|
* rsinfo->setResult. rsinfo->setDesc is set to the tuple description |
302 |
|
|
* that we actually used to build our tuples with, so the caller can |
303 |
|
|
* verify we did what it was expecting. |
304 |
|
|
*/ |
305 |
|
|
rsinfo->setDesc = tupdesc; |
306 |
|
|
MemoryContextSwitchTo(oldcontext); |
307 |
|
|
|
308 |
dpavlin |
31 |
est_cond_delete(cond); |
309 |
|
|
|
310 |
dpavlin |
19 |
if(!est_db_close(db, &ecode)){ |
311 |
|
|
ereport(ERROR, (errcode(ERRCODE_IO_ERROR), |
312 |
|
|
errmsg("est_db_close: %d", ecode), |
313 |
|
|
errdetail(est_err_msg(ecode)))); |
314 |
|
|
} |
315 |
|
|
|
316 |
dpavlin |
1 |
return (Datum) 0; |
317 |
|
|
} |
318 |
|
|
|
319 |
|
|
|
320 |
|
|
/* make text var from attr */ |
321 |
|
|
char *attr2text(ESTDOC *doc, char *attr) { |
322 |
|
|
char *val; |
323 |
|
|
const char *attrval; |
324 |
|
|
int len; |
325 |
dpavlin |
4 |
int attrlen; |
326 |
dpavlin |
1 |
|
327 |
dpavlin |
47 |
if (! doc) return (Datum) NULL; |
328 |
|
|
|
329 |
dpavlin |
56 |
elog(DEBUG1, "doc: %p, attr: %s", doc, attr); |
330 |
dpavlin |
1 |
|
331 |
dpavlin |
4 |
if ( (attrval = est_doc_attr(doc, attr)) && (attrlen = strlen(attrval)) ) { |
332 |
|
|
val = (char *) palloc(attrlen * sizeof(char)); |
333 |
dpavlin |
1 |
} else { |
334 |
|
|
return (Datum) NULL; |
335 |
|
|
} |
336 |
|
|
|
337 |
|
|
len = strlen(attrval); |
338 |
dpavlin |
2 |
elog(DEBUG1, "attr2text(%s) = '%s' %d bytes", attr, attrval, len); |
339 |
dpavlin |
1 |
|
340 |
|
|
len++; |
341 |
|
|
len *= sizeof(char); |
342 |
|
|
|
343 |
|
|
elog(DEBUG2, "palloc(%d)", len); |
344 |
|
|
|
345 |
|
|
val = palloc(len); |
346 |
|
|
|
347 |
|
|
memset(val, 0, len); |
348 |
|
|
strncpy(val, attrval, len); |
349 |
|
|
|
350 |
|
|
elog(DEBUG2, "val=%s", val); |
351 |
|
|
|
352 |
|
|
return val; |
353 |
|
|
} |
354 |
|
|
|
355 |
dpavlin |
41 |
/* |
356 |
|
|
* variation on theme: use node API which doesn't open index on |
357 |
|
|
* every query which is much faster for large indexes |
358 |
|
|
* |
359 |
|
|
*/ |
360 |
|
|
|
361 |
dpavlin |
48 |
/* select * from pgest( */ |
362 |
|
|
#define _arg_node_uri 0 |
363 |
|
|
#define _arg_login 1 |
364 |
|
|
#define _arg_passwd 2 |
365 |
dpavlin |
49 |
#define _arg_depth 3 |
366 |
|
|
#define _arg_query 4 |
367 |
|
|
#define _arg_attr 5 |
368 |
|
|
#define _arg_order 6 |
369 |
|
|
#define _arg_limit 7 |
370 |
|
|
#define _arg_offset 8 |
371 |
|
|
#define _arg_attr_array 9 |
372 |
dpavlin |
48 |
/* as (foo text, ... ); */ |
373 |
|
|
|
374 |
|
|
|
375 |
dpavlin |
41 |
PG_FUNCTION_INFO_V1(pgest_node); |
376 |
|
|
Datum pgest_node(PG_FUNCTION_ARGS) |
377 |
|
|
{ |
378 |
dpavlin |
48 |
ArrayType *attr_arr = PG_GETARG_ARRAYTYPE_P(_arg_attr_array); |
379 |
dpavlin |
41 |
Oid attr_element_type = ARR_ELEMTYPE(attr_arr); |
380 |
|
|
int attr_ndims = ARR_NDIM(attr_arr); |
381 |
|
|
int *attr_dim_counts = ARR_DIMS(attr_arr); |
382 |
|
|
int *attr_dim_lower_bounds = ARR_LBOUND(attr_arr); |
383 |
|
|
int ncols = 0; |
384 |
|
|
int nrows = 0; |
385 |
|
|
int indx[MAXDIM]; |
386 |
|
|
int16 attr_len; |
387 |
|
|
bool attr_byval; |
388 |
|
|
char attr_align; |
389 |
|
|
ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; |
390 |
|
|
AttInMetadata *attinmeta; |
391 |
|
|
TupleDesc tupdesc; |
392 |
|
|
Tuplestorestate *tupstore = NULL; |
393 |
|
|
HeapTuple tuple; |
394 |
|
|
MemoryContext per_query_ctx; |
395 |
|
|
MemoryContext oldcontext; |
396 |
|
|
Datum dvalue; |
397 |
|
|
char **values; |
398 |
|
|
int rsinfo_ncols; |
399 |
|
|
int i, j; |
400 |
|
|
/* estvars */ |
401 |
|
|
ESTNODE *node; |
402 |
|
|
ESTCOND *cond; |
403 |
|
|
ESTNODERES *nres; |
404 |
|
|
ESTRESDOC *rdoc; |
405 |
dpavlin |
57 |
CBMAP *hints; |
406 |
dpavlin |
41 |
int resnum = 0; |
407 |
|
|
int limit = 0; |
408 |
|
|
int offset = 0; |
409 |
dpavlin |
49 |
int depth = 0; |
410 |
dpavlin |
41 |
|
411 |
|
|
char *node_url; |
412 |
|
|
char *user, *passwd; |
413 |
|
|
char *query; |
414 |
|
|
char *attr; |
415 |
|
|
char *order; |
416 |
|
|
|
417 |
|
|
|
418 |
|
|
/* only allow 1D input array */ |
419 |
|
|
if (attr_ndims == 1) |
420 |
|
|
{ |
421 |
|
|
ncols = attr_dim_counts[0]; |
422 |
|
|
} |
423 |
|
|
else |
424 |
|
|
ereport(ERROR, |
425 |
|
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
426 |
|
|
errmsg("invalid input array"), |
427 |
|
|
errdetail("Input array must have 1 dimension"))); |
428 |
|
|
|
429 |
|
|
/* check to see if caller supports us returning a tuplestore */ |
430 |
|
|
if (!rsinfo || !(rsinfo->allowedModes & SFRM_Materialize)) |
431 |
|
|
ereport(ERROR, |
432 |
|
|
(errcode(ERRCODE_SYNTAX_ERROR), |
433 |
|
|
errmsg("materialize mode required, but it is not " \ |
434 |
|
|
"allowed in this context"))); |
435 |
|
|
|
436 |
|
|
/* get info about element type needed to construct the array */ |
437 |
|
|
get_typlenbyvalalign(attr_element_type, &attr_len, &attr_byval, &attr_align); |
438 |
|
|
|
439 |
|
|
/* get the requested return tuple description */ |
440 |
|
|
tupdesc = rsinfo->expectedDesc; |
441 |
|
|
rsinfo_ncols = tupdesc->natts; |
442 |
|
|
|
443 |
|
|
/* |
444 |
|
|
* The requested tuple description better match up with the array |
445 |
|
|
* we were given. |
446 |
|
|
*/ |
447 |
|
|
if (rsinfo_ncols != ncols) |
448 |
|
|
ereport(ERROR, |
449 |
|
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
450 |
|
|
errmsg("invalid input array"), |
451 |
|
|
errdetail("Number of elements in array must match number of query specified columns."))); |
452 |
|
|
|
453 |
|
|
/* OK, use it */ |
454 |
|
|
attinmeta = TupleDescGetAttInMetadata(tupdesc); |
455 |
|
|
|
456 |
|
|
/* Now go to work */ |
457 |
|
|
rsinfo->returnMode = SFRM_Materialize; |
458 |
|
|
|
459 |
|
|
per_query_ctx = fcinfo->flinfo->fn_mcxt; |
460 |
|
|
oldcontext = MemoryContextSwitchTo(per_query_ctx); |
461 |
|
|
|
462 |
|
|
/* initialize our tuplestore */ |
463 |
|
|
tupstore = tuplestore_begin_heap(true, false, SortMem); |
464 |
|
|
|
465 |
|
|
|
466 |
|
|
/* take rest of arguments from function */ |
467 |
|
|
|
468 |
|
|
/* node URL */ |
469 |
dpavlin |
48 |
if (PG_ARGISNULL(_arg_node_uri)) { |
470 |
dpavlin |
41 |
ereport(ERROR, |
471 |
|
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
472 |
|
|
errmsg("node URL can't be null"), |
473 |
|
|
errdetail("Node URL must be valid URL to HyperEstraier node"))); |
474 |
|
|
} |
475 |
dpavlin |
48 |
node_url = _textout(PG_GETARG_TEXT_P(_arg_node_uri)); |
476 |
dpavlin |
41 |
|
477 |
|
|
/* login and password */ |
478 |
dpavlin |
48 |
if (PG_ARGISNULL(_arg_login) || PG_ARGISNULL(_arg_passwd)) { |
479 |
dpavlin |
41 |
ereport(ERROR, |
480 |
|
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
481 |
|
|
errmsg("username and password can't be NULL"), |
482 |
|
|
errdetail("You must specify valid username and password to HyperEstraier node"))); |
483 |
|
|
} |
484 |
dpavlin |
48 |
user = _textout(PG_GETARG_TEXT_P(_arg_login)); |
485 |
|
|
passwd = _textout(PG_GETARG_TEXT_P(_arg_passwd)); |
486 |
dpavlin |
41 |
|
487 |
dpavlin |
49 |
/* depth of search */ |
488 |
|
|
if (PG_ARGISNULL(_arg_depth)) { |
489 |
|
|
depth = 0; |
490 |
|
|
} else { |
491 |
|
|
depth = PG_GETARG_INT32(_arg_depth); |
492 |
|
|
} |
493 |
|
|
|
494 |
dpavlin |
41 |
/* query string */ |
495 |
dpavlin |
48 |
if (PG_ARGISNULL(_arg_query)) { |
496 |
dpavlin |
41 |
query = ""; |
497 |
|
|
} else { |
498 |
dpavlin |
48 |
query = _textout(PG_GETARG_TEXT_P(_arg_query)); |
499 |
dpavlin |
41 |
} |
500 |
|
|
|
501 |
|
|
/* atribute filter */ |
502 |
dpavlin |
48 |
if (PG_ARGISNULL(_arg_attr)) { |
503 |
dpavlin |
41 |
attr = ""; |
504 |
|
|
} else { |
505 |
dpavlin |
48 |
attr = _textout(PG_GETARG_TEXT_P(_arg_attr)); |
506 |
dpavlin |
41 |
} |
507 |
|
|
|
508 |
|
|
/* sort order */ |
509 |
dpavlin |
48 |
if (PG_ARGISNULL(_arg_order)) { |
510 |
dpavlin |
41 |
order = ""; |
511 |
|
|
} else { |
512 |
dpavlin |
48 |
order = _textout(PG_GETARG_TEXT_P(_arg_order)); |
513 |
dpavlin |
41 |
} |
514 |
|
|
|
515 |
|
|
|
516 |
|
|
/* limit */ |
517 |
dpavlin |
48 |
if (PG_ARGISNULL(_arg_limit)) { |
518 |
dpavlin |
41 |
limit = 0; |
519 |
|
|
} else { |
520 |
dpavlin |
48 |
limit = PG_GETARG_INT32(_arg_limit); |
521 |
dpavlin |
41 |
} |
522 |
|
|
|
523 |
|
|
/* offset */ |
524 |
dpavlin |
48 |
if (PG_ARGISNULL(_arg_offset)) { |
525 |
dpavlin |
41 |
offset = 0; |
526 |
|
|
} else { |
527 |
dpavlin |
48 |
offset = PG_GETARG_INT32(_arg_offset); |
528 |
dpavlin |
41 |
} |
529 |
|
|
|
530 |
|
|
/* initialize the network environment */ |
531 |
|
|
if(!est_init_net_env()){ |
532 |
|
|
ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED), |
533 |
|
|
errmsg("pgest_node: can't create network enviroment"))); |
534 |
|
|
} |
535 |
|
|
|
536 |
|
|
/* create the node connection object */ |
537 |
|
|
elog(DEBUG1, "pgest_node: est_node_new(%s) as %s", node_url, user); |
538 |
|
|
node = est_node_new(node_url); |
539 |
|
|
est_node_set_auth(node, user, passwd); |
540 |
|
|
|
541 |
dpavlin |
49 |
elog(DEBUG1, "pgest_node: node: %s (d:%d) query[%s] attr[%s] limit %d offset %d", node_url, depth, query, (PG_ARGISNULL(_arg_attr) ? "NULL" : attr), limit, offset); |
542 |
dpavlin |
41 |
|
543 |
|
|
/* create a search condition object */ |
544 |
|
|
if (!(cond = est_cond_new())) { |
545 |
|
|
ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED), |
546 |
|
|
errmsg("pgest_node: est_cond_new failed"))); |
547 |
|
|
} |
548 |
|
|
|
549 |
|
|
/* set the search phrase to the search condition object */ |
550 |
dpavlin |
48 |
if (! PG_ARGISNULL(_arg_query) && strlen(query) > 0) |
551 |
dpavlin |
41 |
est_cond_set_phrase(cond, query); |
552 |
|
|
|
553 |
|
|
/* minimum valid attribute length is 10: @a STREQ a */ |
554 |
dpavlin |
48 |
if (! PG_ARGISNULL(_arg_attr) && strlen(attr) >= 10) { |
555 |
dpavlin |
41 |
elog(DEBUG1,"attributes: %s", attr); |
556 |
dpavlin |
56 |
cond_add_attr(cond, attr); |
557 |
dpavlin |
41 |
} |
558 |
|
|
|
559 |
|
|
/* set the search phrase to the search condition object */ |
560 |
dpavlin |
48 |
if (! PG_ARGISNULL(_arg_order) && strlen(order) > 0) { |
561 |
dpavlin |
41 |
elog(DEBUG1,"est_cond_set_order(%s)", order); |
562 |
|
|
est_cond_set_order(cond, order); |
563 |
|
|
} |
564 |
|
|
|
565 |
|
|
if (limit) { |
566 |
|
|
elog(DEBUG1,"est_cond_set_max(%d)", limit + offset); |
567 |
|
|
est_cond_set_max(cond, limit + offset); |
568 |
|
|
} |
569 |
|
|
|
570 |
dpavlin |
51 |
if (offset) { |
571 |
|
|
elog(DEBUG1,"est_cond_set_skip(%d)", offset); |
572 |
|
|
est_cond_set_skip(cond, offset); |
573 |
|
|
} |
574 |
|
|
|
575 |
dpavlin |
41 |
/* get the result of search */ |
576 |
dpavlin |
49 |
nres = est_node_search(node, cond, depth); |
577 |
dpavlin |
41 |
|
578 |
dpavlin |
44 |
if (! nres) { |
579 |
|
|
int status = est_node_status(node); |
580 |
|
|
est_cond_delete(cond); |
581 |
|
|
est_node_delete(node); |
582 |
|
|
est_free_net_env(); |
583 |
|
|
ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED), |
584 |
|
|
errmsg("pgest_node: search failed, node status %d", status))); |
585 |
|
|
} |
586 |
|
|
|
587 |
dpavlin |
41 |
/* get number of results */ |
588 |
|
|
resnum = est_noderes_doc_num(nres); |
589 |
|
|
|
590 |
|
|
/* check if results exists */ |
591 |
|
|
if ( 0 == resnum ) { |
592 |
|
|
elog(INFO, "pgest_node: no results for: %s", query ); |
593 |
|
|
} |
594 |
|
|
|
595 |
|
|
/* total number of tuples to be returned */ |
596 |
|
|
if (limit && limit < resnum) { |
597 |
|
|
nrows = limit; |
598 |
|
|
} else { |
599 |
dpavlin |
80 |
nrows = resnum; |
600 |
dpavlin |
41 |
} |
601 |
|
|
|
602 |
dpavlin |
57 |
/* get hints */ |
603 |
|
|
hints = est_noderes_hints(nres); |
604 |
|
|
|
605 |
dpavlin |
41 |
elog(DEBUG1, "pgest_node: found %d hits for %s", resnum, query); |
606 |
|
|
|
607 |
|
|
|
608 |
|
|
values = (char **) palloc(ncols * sizeof(char *)); |
609 |
|
|
|
610 |
|
|
for (i = 0; i < nrows; i++) |
611 |
|
|
{ |
612 |
|
|
|
613 |
|
|
/* get result from estraier */ |
614 |
dpavlin |
51 |
if (! ( rdoc = est_noderes_get_doc(nres, i) )) { |
615 |
dpavlin |
47 |
elog(INFO, "pgest_node: can't find result %d", i + offset); |
616 |
dpavlin |
41 |
} else { |
617 |
|
|
elog(DEBUG1, "URI: %s\n Title: %s\n", |
618 |
|
|
est_resdoc_attr(rdoc, "@uri"), |
619 |
|
|
est_resdoc_attr(rdoc, "@title") |
620 |
|
|
); |
621 |
|
|
} |
622 |
|
|
|
623 |
|
|
/* iterate over results */ |
624 |
|
|
for (j = 0; j < ncols; j++) |
625 |
|
|
{ |
626 |
|
|
bool isnull; |
627 |
dpavlin |
57 |
char *attr; /* current attribute name */ |
628 |
|
|
char *hint; /* position of current hint in attribute */ |
629 |
|
|
char *hint_val; |
630 |
dpavlin |
41 |
|
631 |
|
|
/* array value of this position */ |
632 |
|
|
indx[0] = j + attr_dim_lower_bounds[0]; |
633 |
|
|
|
634 |
|
|
dvalue = array_ref(attr_arr, attr_ndims, indx, -1, attr_len, attr_byval, attr_align, &isnull); |
635 |
dpavlin |
57 |
attr = (char *)DirectFunctionCall1(textout, dvalue); |
636 |
dpavlin |
41 |
|
637 |
dpavlin |
57 |
if (!isnull && (hint = strstr(attr, HINTS_PREFIX)) != NULL) { |
638 |
|
|
/* skip HINTS. prefix */ |
639 |
|
|
hint += strlen(HINTS_PREFIX); |
640 |
|
|
|
641 |
dpavlin |
72 |
hint_val = (char *)cbmapget(hints, hint, -1, NULL); |
642 |
dpavlin |
57 |
elog(DEBUG2, "hint %s = %s", hint, hint_val); |
643 |
|
|
|
644 |
|
|
if (hint_val != NULL) { |
645 |
|
|
values[j] = DatumGetCString( hint_val ); |
646 |
|
|
} else { |
647 |
|
|
elog(INFO, "can't get hint in results: %s", hint); |
648 |
|
|
values[j] = NULL; |
649 |
|
|
} |
650 |
|
|
} else if (!isnull && rdoc) |
651 |
|
|
values[j] = DatumGetCString( node_attr2text(rdoc, attr) ); |
652 |
dpavlin |
41 |
else |
653 |
|
|
values[j] = NULL; |
654 |
|
|
} |
655 |
|
|
/* construct the tuple */ |
656 |
|
|
tuple = BuildTupleFromCStrings(attinmeta, values); |
657 |
|
|
|
658 |
|
|
/* now store it */ |
659 |
|
|
tuplestore_puttuple(tupstore, tuple); |
660 |
|
|
|
661 |
|
|
} |
662 |
|
|
|
663 |
|
|
tuplestore_donestoring(tupstore); |
664 |
|
|
rsinfo->setResult = tupstore; |
665 |
|
|
|
666 |
|
|
/* |
667 |
|
|
* SFRM_Materialize mode expects us to return a NULL Datum. The actual |
668 |
|
|
* tuples are in our tuplestore and passed back through |
669 |
|
|
* rsinfo->setResult. rsinfo->setDesc is set to the tuple description |
670 |
|
|
* that we actually used to build our tuples with, so the caller can |
671 |
|
|
* verify we did what it was expecting. |
672 |
|
|
*/ |
673 |
|
|
rsinfo->setDesc = tupdesc; |
674 |
|
|
MemoryContextSwitchTo(oldcontext); |
675 |
|
|
|
676 |
|
|
/* delete the node result object */ |
677 |
|
|
est_noderes_delete(nres); |
678 |
|
|
|
679 |
|
|
/* destroy the search condition object */ |
680 |
|
|
est_cond_delete(cond); |
681 |
|
|
|
682 |
|
|
/* destroy the node object */ |
683 |
|
|
est_node_delete(node); |
684 |
|
|
|
685 |
|
|
/* free the networking environment */ |
686 |
|
|
est_free_net_env(); |
687 |
|
|
|
688 |
|
|
return (Datum) 0; |
689 |
|
|
} |
690 |
|
|
|
691 |
|
|
/* make text var from node attr */ |
692 |
|
|
char *node_attr2text(ESTRESDOC *rdoc, char *attr) { |
693 |
|
|
char *val; |
694 |
|
|
const char *attrval; |
695 |
|
|
int len; |
696 |
|
|
int attrlen; |
697 |
|
|
|
698 |
dpavlin |
47 |
if (! rdoc) return (Datum) NULL; |
699 |
|
|
|
700 |
dpavlin |
56 |
elog(DEBUG1, "doc: %p, attr: %s", rdoc, attr); |
701 |
dpavlin |
41 |
|
702 |
|
|
if ( (attrval = est_resdoc_attr(rdoc, attr)) && (attrlen = strlen(attrval)) ) { |
703 |
|
|
val = (char *) palloc(attrlen * sizeof(char)); |
704 |
|
|
} else { |
705 |
|
|
return (Datum) NULL; |
706 |
|
|
} |
707 |
|
|
|
708 |
|
|
len = strlen(attrval); |
709 |
|
|
elog(DEBUG1, "node_attr2text(%s) = '%s' %d bytes", attr, attrval, len); |
710 |
|
|
|
711 |
|
|
len++; |
712 |
|
|
len *= sizeof(char); |
713 |
|
|
|
714 |
|
|
elog(DEBUG2, "palloc(%d)", len); |
715 |
|
|
|
716 |
|
|
val = palloc(len); |
717 |
|
|
|
718 |
|
|
memset(val, 0, len); |
719 |
|
|
strncpy(val, attrval, len); |
720 |
|
|
|
721 |
|
|
elog(DEBUG2, "val=%s", val); |
722 |
|
|
|
723 |
|
|
return val; |
724 |
|
|
} |
725 |
|
|
|
726 |
dpavlin |
56 |
/* parse attributes and add them to confition */ |
727 |
|
|
void cond_add_attr(ESTCOND *cond, char *attr) { |
728 |
|
|
char *next; |
729 |
|
|
char *curr_attr; |
730 |
|
|
while ( strlen(attr) > 0 ) { |
731 |
dpavlin |
73 |
printf("len [%s] = %zd\n", attr, strlen(attr)); |
732 |
dpavlin |
56 |
if ((next = strstr(attr, ATTR_DELIMITER)) != NULL) { |
733 |
|
|
curr_attr = palloc( next - attr + 1 ); |
734 |
|
|
memcpy(curr_attr, attr, next-attr); |
735 |
|
|
curr_attr[next-attr] = '\0'; |
736 |
|
|
next += strlen(ATTR_DELIMITER); |
737 |
|
|
} else { |
738 |
|
|
next = ""; |
739 |
|
|
curr_attr = attr; |
740 |
|
|
} |
741 |
|
|
elog(DEBUG1, "est_cond_add_attr(%s)", curr_attr); |
742 |
|
|
est_cond_add_attr(cond, curr_attr); |
743 |
|
|
attr = next; |
744 |
|
|
} |
745 |
|
|
} |
746 |
dpavlin |
61 |
|
747 |
|
|
/* trigger to keep data in Hyper Estraier index up-to-date */ |
748 |
|
|
/* CREATE FUNCTION pgest_trigger() RETURNS TRIGGER AS ... */ |
749 |
|
|
|
750 |
|
|
/* |
751 |
|
|
* UPDATE, INSERT and DELETE triggers are like this: |
752 |
|
|
|
753 |
|
|
CREATE TRIGGER pgest_trigger_update AFTER UPDATE |
754 |
|
|
ON table_name FOR EACH ROW |
755 |
|
|
EXECUTE PROCEDURE pgest_trigger('http://localhost:1978/node/trivia','admin','admin', |
756 |
|
|
'name_of_pk', 'column', 'another_column', 'and_so_on' |
757 |
|
|
) |
758 |
|
|
|
759 |
|
|
*/ |
760 |
|
|
|
761 |
|
|
PG_FUNCTION_INFO_V1(pgest_trigger); |
762 |
|
|
Datum pgest_trigger(PG_FUNCTION_ARGS) { |
763 |
|
|
|
764 |
|
|
TriggerData *data; |
765 |
|
|
TupleDesc tupdesc; |
766 |
|
|
HeapTuple ret; |
767 |
|
|
|
768 |
|
|
char **args; |
769 |
|
|
char *keycol = NULL; |
770 |
|
|
char *key = NULL; |
771 |
|
|
char *col_data = NULL; |
772 |
|
|
int knumber; |
773 |
|
|
int i; |
774 |
|
|
int create_doc = 0; |
775 |
dpavlin |
79 |
int edit_doc = 0; |
776 |
dpavlin |
61 |
|
777 |
|
|
ESTNODE *node; |
778 |
|
|
ESTDOC *doc; |
779 |
|
|
|
780 |
|
|
|
781 |
|
|
|
782 |
|
|
if (! CALLED_AS_TRIGGER(fcinfo)) { |
783 |
|
|
elog(ERROR, "pgest_trigger() must be called as a trigger"); |
784 |
|
|
} |
785 |
|
|
|
786 |
|
|
data = (TriggerData *) fcinfo->context; |
787 |
|
|
|
788 |
|
|
if (data->tg_trigger->tgnargs < 5) |
789 |
|
|
elog(ERROR, "pgest_trigger() requires at least 5 parameters ('http://localhost:1978/node/trivia', 'user', 'passwd', 'pk_column', 'column', ... )"); |
790 |
|
|
|
791 |
|
|
args = data->tg_trigger->tgargs; |
792 |
|
|
keycol = args[3]; |
793 |
|
|
|
794 |
|
|
tupdesc = data->tg_relation->rd_att; |
795 |
|
|
|
796 |
|
|
knumber = SPI_fnumber(tupdesc, keycol); |
797 |
|
|
key = SPI_getvalue(data->tg_trigtuple, tupdesc, knumber); |
798 |
|
|
|
799 |
|
|
|
800 |
|
|
/* initialize the network environment */ |
801 |
|
|
if( ! est_init_net_env() ) |
802 |
|
|
elog(ERROR, "pgest_trigger: network is unavailable\n"); |
803 |
|
|
|
804 |
|
|
/* create and configure the node connection object */ |
805 |
|
|
node = est_node_new( args[0] ); |
806 |
|
|
est_node_set_auth(node, args[1], args[2]); |
807 |
|
|
|
808 |
|
|
|
809 |
|
|
if (TRIGGER_FIRED_BY_INSERT(data->tg_event)) { |
810 |
|
|
/* There is no old data */ |
811 |
|
|
ret = data->tg_trigtuple; |
812 |
|
|
|
813 |
|
|
create_doc++; |
814 |
|
|
|
815 |
|
|
} else if (TRIGGER_FIRED_BY_UPDATE(data->tg_event)) { |
816 |
|
|
ret = data->tg_newtuple; |
817 |
|
|
|
818 |
dpavlin |
79 |
edit_doc++; |
819 |
dpavlin |
61 |
|
820 |
|
|
} else if (TRIGGER_FIRED_BY_DELETE(data->tg_event)) { |
821 |
|
|
/* There is no new data */ |
822 |
|
|
ret = data->tg_trigtuple; |
823 |
|
|
|
824 |
|
|
if (! est_node_out_doc_by_uri(node, key) ) |
825 |
|
|
elog(ERROR, "est_node_doc_by_uri(%s): %d\n", key, est_node_status(node)); |
826 |
|
|
|
827 |
|
|
} else { |
828 |
|
|
elog(ERROR, "pgest_trigger() not called from INSERT/UPDATE/DELETE"); |
829 |
|
|
} |
830 |
|
|
|
831 |
dpavlin |
79 |
if ( create_doc || edit_doc ) { |
832 |
dpavlin |
61 |
|
833 |
dpavlin |
79 |
if ( create_doc ) { |
834 |
|
|
/* create a document object */ |
835 |
|
|
doc = est_doc_new(); |
836 |
|
|
est_doc_add_attr(doc, "@uri", key); |
837 |
dpavlin |
61 |
|
838 |
dpavlin |
79 |
elog(DEBUG1, "est_doc_new @uri=%s", key); |
839 |
|
|
} else { |
840 |
|
|
/* edit existing document */ |
841 |
|
|
doc = est_node_get_doc_by_uri(node, key); |
842 |
|
|
if (doc == NULL) |
843 |
|
|
elog(ERROR, "est_node_get_doc_by_uri(%s): %d\n", key, est_node_status(node)); |
844 |
|
|
elog(DEBUG1, "est_node_get_doc_by_uri(%s)", key); |
845 |
|
|
} |
846 |
dpavlin |
61 |
|
847 |
|
|
for( i = 4; i < data->tg_trigger->tgnargs; i++ ) { |
848 |
|
|
|
849 |
|
|
col_data = SPI_getvalue(ret, tupdesc, SPI_fnumber(tupdesc, args[i])); |
850 |
|
|
|
851 |
|
|
if (data) { |
852 |
|
|
elog(DEBUG1, " + %s = %s", args[i], col_data); |
853 |
|
|
est_doc_add_attr(doc, args[i], col_data); |
854 |
|
|
est_doc_add_text(doc, col_data); |
855 |
|
|
} |
856 |
|
|
|
857 |
|
|
} |
858 |
|
|
|
859 |
dpavlin |
79 |
if ( edit_doc ) { |
860 |
|
|
/* update existing document */ |
861 |
|
|
if( ! est_node_edit_doc(node, doc) ) |
862 |
|
|
elog(ERROR, "est_node_edit_doc: %d\n", est_node_status(node)); |
863 |
|
|
} else { |
864 |
|
|
/* register the document object to the node */ |
865 |
|
|
if( ! est_node_put_doc(node, doc) ) |
866 |
|
|
elog(ERROR, "est_node_put_doc: %d\n", est_node_status(node)); |
867 |
|
|
} |
868 |
dpavlin |
61 |
|
869 |
|
|
/* destroy the document object */ |
870 |
|
|
est_doc_delete(doc); |
871 |
|
|
|
872 |
|
|
} |
873 |
|
|
|
874 |
|
|
/* destroy the node object */ |
875 |
|
|
est_node_delete(node); |
876 |
|
|
/* free the networking environment */ |
877 |
|
|
est_free_net_env(); |
878 |
|
|
|
879 |
|
|
|
880 |
|
|
return PointerGetDatum(ret); |
881 |
|
|
} |
882 |
|
|
|